ADDITIONAL FATIGUE TESTS OF HYBRID PLATE

GIRDERS UNDER PURE BENDING MOMENT

by

Y. Kurobane

D. J. Fielding

A, A. Toprac

Research Report Number 96-1

Research Project Number 3-5-66-96

Fatigue Strength of Hybrid Plate Girders under Shear

Conducted for

Texas Highway Department

In Cooperation with the

U. S. Department of Transportation

Federal Highway Administration

Bureau of Public Roads

by

CENTER FOR HIGHWAY RESEARCH

THE UNIVERSITY OF TEXAS

AUSTIN, TEXAS

May 1967

ACKNOW LEDGMENTS

This investigation is part of the hybrid steel plate girder

research being conducted at The University of Texas at Austin,

Center for Highway Research under the administrative direction

of Dr. John J. McKetta. The research was sponsored by the

Texas Highway Department in cooperation with the U. S. Depart­

ment of Transportation, Bureau of Public Roads.

The authors wish to thank Mr. Wayne Henneberger and Mr.

Van Der Straten for their interest in the test program. Assistance

during the testing program was provided by Messrs. H. S. Lew,

Dale G. Eyre and Robert Garcia.

The opinions, findings, and conclusions expressed in this

publication are those of the authors and not necessarily those of

the Bureau of Public Roads.

The present report is the first of three under Project 3-5-66-96

"Fatigue Strength of Hybrid Plate Girders Under Shear." The

second report under this project is entitled IIHybrid Plate Girders

Under Combined Bending and Shear, II and will be issued shortly.

The third, and final report will pertain to the fatigue strength of

hybrid plate girders under constant moment.

ABSTRACT

Six fatigue tests were conducted on welded hybrid plate

girders. The flange steel was A44l and the web steel was

A36. Web slenderness ratios of 144 and 192 were investigated.

A sequential test program was devised to determine the maxi­

mum permissible stres s range that will give a life of two million

cycles.

The objectives of this investigation were to obtain fatigue

data for hybrid plate girders and to determine the fatigue strength

at two million cycles.

TABLE OF CONTENTS

Page

Abstract iii

l. Introduction 1

2. Test Program and Specimens 4

3. Test Set- Up, Measurements, and Testing Procedure 7

4. Fa tigue Cracks 10

5. Discussion of Test Results 13

6. Summary 16

References 18

Tables 19

Figures 22

1

1. INTRODUCTION

Since 1961 both static and fatigue behavior of hybrid plate

girders have been studied at The University of Texas. Prior to

the investigation described here, both static - and fatigue - test

results from a number of research projects have been reported.

For these tests the specimens consisted of ASTM A5l4 steel flanges

and A36 steel webs. (I), (2), (3) Since there was no data available

on the fatigue strength of plate girders with A44l flanges and A36

webs, an investigation of this type of hybrid sections was thought

to be useful.

Hybrid girders with A5l4 flanges and A36 webs, when sub-

jected to cyclic loads of pure bending type. developed three types

of fatigue cracks, (2), (3) referred to as Types I, 2, and 3.

Type 1 cracks are those which occurred in the compression

side of the web plate along the toe of the flange-to-web fillet weld.

The thin web plates of hybrid girders deform laterally as the load

increases, producing larger deflection in the compression side of

the web panels than in the tension side. It was inferred that the

fluctuating stress at the compression edge of the web due to this web

movement is pertinent to this type of crack formation. In general

Type 1 cracks were observed in the girders with very slender webs

(web slenderness ratio greater than 192).

Type 2 cracks are those which initiate at the stiffener-to-web

boundary near the cut-off end of the stiffener. a few inches above the

tension flange. The major cause for this crack initiation was consider-

ed to be the tensile primary bending fiber stres s coupled with a stress

2

concentration due to the abrupt termination of vertical stiffeners

near the tension flange. Type 2 cracks were found in most

groups of test girders, regardless of the web thickness. These

cracks extend both upward and downward along the toe of the

web-to-stiffener fillet weld. When this crack reaches the ten-

sion flange, it propagates into it and eventually fractures the

flange. The propagation of these cracks is much faster than the

propagation of Type 1 cracks.

Type 3 cracks are those which occurred in the tension

flange or in the web-to-flange fillet weld at points where notches

existed. These notches were produced by defective welding and

cutting ope ra tions.

In the tests reported previously (3) no cracks in the bend­

ing panels were found within two million cycles if the applied

maximum stress in the flanges was less than 30 ksi and the stress

range was less than 15 ksi. For the specimens which were sub­

jected to a flange stress of more than 40 ksi, the three types of

cracks described above were obs erved.

The tests presented here are an extension of the investiga­

tion reported in Ref. 3. The specific purpose of this project is to

secure test data on the fatigue behavior of hybrid girders under

pure bending when the section is composed of A44l steel flanges

and an A36 web. In this test series the maximum flange stress

was kept at 30 ksi (60 percent of the specified yield point of the flange

material), and the stress range was varied (see Table 3).

Information was sought as to the stres s range within which the

bending panels can survive approximately two million cycles.

3

This number of cycles was chosen because this was considered

to be the largest number of cycles of maximum load that any

member in a bridge would ever experience during its service life.

4

2. TEST PROGRAM AND SPECIMENS

To study the fatigue strength of the girders at two million

cycles a sequential test plan as described below was adopted.

The first specimen was tested with extreme fiber stres s

fluctuating between 10 ksi and 30 ksi. If the specimen failed be­

fore two million cycles the next specimen was to be tested with a

stress range 5 ksi less than the previous stress range, keeping

the maximum stress constant at 30 ksi. If the first specimen did

not fail before two million cycles, the second specimen was to be

tested with a stress range 5 ksi greater than the previous one.

If the second specimen failed, the stress range was de­

creased by 5 ksi; if the second specimen did not fail, the third

specimen was tested at the same stress level as before to make

sure that another specimen with the same stress range would not

fail within two million cycles. The maximum stres s was kept

at 30 ksi for all the specimens. This proposed sequence of fati­

gue testing is shown in Fig. 1.

Six specirn.ens were test,~d of which three had 3/16 inch

and three had 1/4 inch web thickness. This gave slendernes s

ratios of 192 and 144, respectively. These web slenderness ratios

are above and below the maximum ratio of 170 specified by the

AASHO bridge specifications for A 7 steel. The objective here was

to use practical ratios that are encountered in bridge design.

All the specimens had the same dimensions except the web

thickness. The web was 36 inches deep and was stiffened by pairs

of transverse stiffeners 36 inches apart giving a panel-aspect ratio

(ratio of panel width to depth) of 1. O. The stiffeners were cut

short two inches above the tension flange to reduce the possi­

bility of premature fatigue damage to the tension flange. Both

5

the tension and compression flanges had the same dimensions, and

had uniform width and thickness along the length of the specimens.

The actual measured dimensions of each specimen are shown in

Table 1, along with the nominal dimensions.

Figure 2 shows the specimen configuration. The specimens

had two central test panels, a narrow panel on each end of the test

section, and three panels between the loading jacks and the supports.

The purpose of having the narrow panel between the load point and

the test panel was to provide a transition section to reduce the local

effect of load application. Automatic submerged arc welding 'was

used at the flange to web juncture resulting in a full-penetration

weld. For all other welds E 7018 electrodes were used.

Each specimen was numbered with respect to the thickness

of the web, the minimum and maximum stresses, and the test series

designation. For example, in 3l030H the first number designates

the thickness of the web in multiples of 1/16 inch, the next four

numbers designate the stress range with the first two as the mini­

mum stress and the last two as the maximum stress. The letter H

indicated this series of tests.

For the specimens, all plates of the same thickness were

taken from the same heat. Both the chemical and physical properties

of the flanges and the webs are pres ented in Table 2. The physical

6

properties in this table were measured by means of standard tensile

tests on coupons from the same plates used in the specimens. It

should be noted that the yield stresses given in Table 2 were obtained

at zero strain rate and are referred to as static yield stresses.

7

3. TEST SET-UP, MEASUREMENTS, AND TESTING PROCEDURE

The girders were tested in a simply supported condition

with two equal concentrated loads symmetrically applied with

respect to the center line. The test panels, therefore, were

subjected to constant bending moment (Fig. 2). A hydraulic pul­

sator and hydraulic jacks of 120 kips capacity were used to apply

the cyclic loads. To prevent tilting of the girders, a sufficient

number of lateral braces, designed to allow only vertical move­

ments, were used.

In order to observe the general behavior of the girders,

vertical deflections were measured at the center, at the points of

load application, and at the supports, using dial gages. Lateral

web deflections were measured at 3 inch by 3 inch grid points as

shown in Fig. 3. These latter measurements were made using

a special rig with a dial gage attached to a movable block that could

be positioned at any desired level (Fig. 4). Electrical resistance­

strain gages were mounted on the flanges and on both sides of the

web to check the longitudinal strain under the applied bending

moment.

During the fatigue test, a slip-deflection gage was placed

at the center of a specimen to obtain an indication of the loss of

girder stiffness as a result of crack development.

Preceding the fatigue test of each girder, a static test

was performed. Load was increased up to the maximum load that

was to be applied in the fatigue test, and then removed. At the

maximum and minimum stress levels and zero load, the measurements

8

mentioned above were made. Typical curves for vertical deflec-

tions, lateral web deflections, and strains in the flanges and the

webs are shown in Figs. 5, 6, and 7, respectively. Web deflections

at the center line of two typical test panels, with 1/4 inch and 3/16

inch web respectively are shown in Fig. 6. For the 1/4 inch web,

the deflections changed very little up to the maximum test load P . max

For the 3/16 inch web the deflections increased between P . and P • mln max

The reason for this behavior is that P for the 1/4 inch web is less max

than P (the load to cause web buckling); whereas, for the 3/16 inch cr

web P is larger than P . The actual ratios of P /p are max cr max cr

indicated on Fig. 6. When the maximum test load had been reached,

a thorough inspection of the entire girder was made. All yield lines

were noted by observing cracks in the white-washed surfaces. Welds

were carefully examined visually with the aid of a magnifying glass to

see that no cracks were present.

The maximum and minimum stresses for the fatigue testing

were computed by beam theory based upon the actual dimensions.

All test loads used for the fatigue test were determined from these

stresses and corrections were made for the effect of the inertia force

on the dynamic response of the girder. The fatigue test followed

immediately after the static-load test. The pulsating loads were

applied at a rate of about 250 to 300 cycles per minute. Throughout

the fatigue test visual inspections of all welds and heat affected areas

(due to welding) were made with the aid of a magnifying glass at 3-

hour intervals (about 54, 000 cycles). However, more frequent

observations were made after the first crack was noted. Slip-gage

readings were recorded at each inspection period.

9

To observe crack propagation, the growth of cracks was

marked and measured at each inspection pe riod. The initiation and

growth of the first crack, as well as of any other crack forming

later, are described in the following section. If cracks were found

outside the test section, fatigue testing was stopped temporarily and

the cracks were repaired by first gouging out the cracks by the

"arc-air" method and then depositing fresh weld. Following the

repair, testing was resumed until two million cycles were attained,

unless either a detrimental reduction of the girder stiffness was noted

due to the fracture of the tension flange outside the test panels or

cracks were found in the test panels. Whenever testing was stopped,

the girder was loaded with the maximum test load so that cracks

would be more easily seen and so that a check could be made to insure

that no cracks had been overlooked.

10

4. FATIGUE CRACKS

All cracks found in this series of tests were Type 2

cracks, namely, those occurring along the toe of the web-to­

stiffener fillet weld near the cut-off end of the stiffener. The

fatigue cracks discussed here are only those observed inside the

test section, along the three stiffeners nearest the middle of the

span. In order to avoid confusion, the cracks appearing outside

the test section are not discussed here, but they are included in

the figures showing crack locations.

Testing of 3/16-inch Web Specimens

Specimens 31030H (Fig. 8) - According to the test sequence

adopted, the minimum and maximum stresses in the flanges for

this test were 10 ksi and 30 ksi, respectively. The first crack

was found along the toe of the web-to-stiffener weld at 1,842,000

cycles. This crack was 1 inch long, extended vertically above

the cut-off end of the stiffener, and was observed on one side only.

At 1,868,000 cycles, the second crack was found along another

stiffeher. This crack was 1/2 inch long extending upward from

the cut-off end and was observed on one side only. The testing

was discontinued at 2,015,000 cycles. The cracks had propagated

vertically; the first crack was 2 3/16 inches long reaching a point

1 1/ 4 inches above the tension flange; the second crack was 1/2

inch long reaching a point 1 7/8 inches above the tension flange.

Both of them were seen only on one side of the girder.

Specimen 3l530H (1) - Since the first specimen developed

fatigue cracks within two million cycles, the minimum and maximum

11

stresses for the second specimen were 15 ksi and 30 ksi. No

cracks were found within two million cycles. The testing was

terminated at 2, 941, 000 cycles.

Specimen 3l530H (2) - Since no cracks were found within

two million cycles in the second specimen, the third specimen

was tested at the same stress level as specimen 3l530H (1).

Again no cracks were found within two million cycles. Testing

was terminated at 2,360,000 cycles.

Testing of 1/4-inch Web Specimens

Specimen 4l030H (Fig. 9) - According to the test program,

the minimum and maximum stresses in the flanges of this speci­

men were 10 ksi and 30 ksi, respectively. No cracks were found

within two million cycles. Testing was stopped at 2,041,000

cycles and final inspection was made under a static load to be

certain that no cracks were present. On this occasion, a crack

was found along the toe of the web-to-stiffener fillet weld extend­

ing from the cut-off end of the stiffener. It was 2 inches long and

was seen on one side only.

Specimen 40530H (1) (Fig. 10) - Since no cracks were found

within two million cycles in the first specimen, the stress range

was increased. The minimum and maximum stresses for this

spcimen were 5 ksi and 30 ksi, respectively. A crack similar

to Type 2 cracks appeared along the stiffener outside the test

section at 577,000 cycles. The crack was repaired three times,

but each time it reappeared at the repaired area. The test was

finally terminated at 888, 000 cycles. However, during the final

12

inspection at 888,000 cycles, a crack was observed within the

test area. It was 1 1/4 inches long extending upward from the cut­

off end of the stiffener, and was observed only on one side of the

girder.

Specimen 40530H (2) (Fig. 11) - Although the second speci­

men developed a fatigue crack in the test section within two million

cycles, the third specimen was tested at the same stress level

as the second specimen (Specimen 40530H (1)) as a check on the

results. Two cracks were found along the two stiffeners at

862,000 cycles. One was 6 1/2 inches long extending along the toe

of the weld and reaching a point 1/2 inch above the tension flange.

The other was 2 inches long extending along the toe of the weld

above the cut-off end of the stiffener. Both of tern were observed only

on one side of the girder. The testing was stopped at 934,000

cycles because a crack which occurred outside the test section

reached the tension flange.

A summary of the fatigue test results is shown in Table 3.

13

5. DISCUSSION OF TEST RESULTS

It was noticed in a previous investigation (3) that Type 2 cracks

always propagated into the tension flange as the number of the applied

cycles of loads increased, finally resulting in complete failure of

girders due to fracture of the tension flange. Therefore, the initia­

tion of Type 2 cracks is regarded in this study as fatigue failure for

a girder.

Flexural membrane (fiber) stresses in the web around the cut-

off end of the stiffener can be estimated by simple beam theory. As

shown in Fig. 7 the strains measured by strain gages and the strains

computed by beam theory show some difference. A more thorough

investigation (4) of the membrane stresses in the webs near the

vertical stiffeners gave the conclusion that the difference is small.

These stresses are of course different from the stresses present at the

toes of the welds where the Type 2 cracks appeared, because of the

localized stres s disturbance due to welding.

Figure 12 shows the relationship between the extreme fiber

stress range in the flange (right hand scale) and the number of cycles

when the first crack developed in all tests of this series. In the tests

of series A and B reported previously, (3) Type 2 cracks developed

regardless of the web thickness, and the S-N diagram indicated a

curve which is typical of S-N curves in the fatigue of metal structures.

The stress in the flange is proportional to the membrane stress in the

web regardless of the web thickness. Therefore, it is considered that

the flexural membrane stres s is one of the important factors influenc-

ing the development of Type 2 cracks.

14

The membrane stress in the web at the cut-off end of the stiffener

was about 86 percent of the stress in the flange according to beam

theory. The stress in the web at this point was 26.0 ksi when the

stress in the flange was 30 ksi. The stress range of the web at this

point is indicated in Fig. 12 as the ordinate in the stress-range scale

to the left. The configuration and the state of stress around the area

at which the Type 2 cracks were found seem to be fairly similar to

those of a plate with fillet-welded attachments as shown in Fig. 13.

According to Fig. 8.4 of Ref. 5 the median fatigue strength of the

plate with the fillet-welded attachments at two million cycles is about

26 ksi and 10 ksi in terms of the maximum and the minimum stresses.

This fatigue strength is close to the stress level of the web-membrane

stress at the cut-off ends of the stiffen~rs where cracks were found

at about two million cycles (Fig. 12). The S-N Curve in Fig. 12

was obtained from Ref. 5 assuming a linear log S versus log N

relationship between 105

and 2 x 106

cycles. There are many factors

which can influence the occurrence of Type 2 cracks. A comparison

which was made in the above reference with a simple tension specimen

considered only the effect of the fiber stresses in the web. More in-

vestigation of the factors that influence the initiation of Type 2 cracks

is neces sary.

A sequential test as described in Section 2 was adopted to de-

termine the fatigue strength of the hybrid girders under pure bending

moment at two million cycles. The maximum stress in the flanges

was kept at 30 ksi for all the specimens and the stress range in the

flanges was changed depending on whether or not the preceding

15

specimen failed within two million cycles. The test results are shown

in Fig. 14, where 0 indicates no failure and X indicates failure. The

available test results are still too few to estimate the fatigue strength

for a certain percentage of survival at two million cycles from the

statistical viewpoint. According to the S-N relationship shown in Fig.

12, however, the approximate fatigue strength (at 2 x 106

cycles) is

considered to be about 20 ksi flange stress range for a ITlaxiITlum flange

stress of 30 ksi. The results of these tests are shown again in Fig. 15

with the S-N Curve presented in Ref. 3. The results of this investiga-

tion agree quite well with the previous tests.

16

6. SUMMARY

A series of fatigue tests of hybrid plate girders under pure bend­

ing moment was performed. ASTM A44l steel was used for the

flanges and A36 steel was used for the webs. Two groups of girders

(three specimens for each group) were tested. These two groups

differ only in the web thickness; the one had a web slenderness ratio

of 144, and the other had a web slenderness ratio of 192.

Although the objective of this investigation was to compile

data on the fatigue strength of hybrid girders, special emphasis

was placed on estimating the fatigue strength of the bending panels

at two million cycles. To estimate the fatigue strength at two

million cycles, a sequential series of response tests was adopted.

The maximum flange stress was kept at 30 ksi (60 percent of

specified yield point of A44l steel) and the stress range was

varied.

For all the specimens tested, the fatigue cracks occurred in

the web along the toe of the web-to-stiffener fillet weld near the cut­

off end of the stiffeners. It was noted that the initiation of this type

of crack was related to the tensile fiber stress range in the web.

The difference in web thicknes s apparently was not significant

in the tests carried out. The stress magnitude at the points where

the cracks appeared was found to be comparable with the median

fatigue strength of tension specimens with fillet-welded attachments.

For the bending panels tested, the approximate fatigue strength at

two million cycles was found to be a stres s range of 20 ksi for a

17

maximum stress of 30 ksi. However, more tests are considered

necessary to establish the fatigue strength more precisely and to

carry out a statistical analysis. The S-N curve presented in Fig.

15 provides a slightly conservative estimate of the fatigue strength

of the girders in this test series.

18

REFERENCES

(1) A. A. Toprac and R. A. Engler, IIPlate Girders with High-Strength Steel Flanges and Carbon Steel Webs, II 1961 Proceedings, National Engineering Conference AISC, New York, N. Y.

(2) A. A. Toprac, IIFatigue Strength of Hybrid Plate Girders, II S.F .. R. L. Report No. 04-64, The University of Texas, (1964).

(3) H. S. Lew and A. A. Toprac, IIFatigue Strength of Hybrid Plate Girders Under Constant Moment, II High­way Research Record No. 167, Highway Research Board, Washington, D. C., 1967.

(4) B. T. Yen and J. A. Mueller, "Fatigue Tests of Large­size Welded Plate Girder, II Fritz. Engrg. Lab. Report, 303.10 (June 1966),

(5) W. H. Munse, IIFatigue of Welded Steel Structure, II Welding Research Council (1964).

Nominal Dimensions Web Measured Dimensions Moment of

Specimen No. FIg. Wdth. X Web Depth X Slenderness FIg. Wdth. X Web Depth X Inertia FIg. Th. Web Th. FIg. Th. Web Th.

(in. 4) Ratio

31030 H 8t! x 1/2t! 36!t X 3/16!t 192 8.003 11 x 0.5097 11 36. Off x 0.1972 11 3485

31530 H (1) II 11 11 8.020 11 X O. 51 00t! II 3493

31530 H (2) t! II II 7. 995!t X 0.5083" II 3475

41030 H " 36" x 1/4 " 144 8,007" x 0.5093 11 36.011 x 0.2617 11 3735

40530 H (1) II II " 7.983" x O. 5072!t " 3716

40530 H (2) II " " 8,006" x 0.5112" " 3745

TABLE 1 Cross Sectional Dimensions of Test Panels

PLATE 1/2 11 (A441) 1/4"(A36)

Thickness (in.) 0.514 0.261

Static Y. P. (ksi) 52. 7 43.3

Tensile Str. (ksi) 74.2 66. 3

Elong. (% in 8 in. ) 24 29

Chern. Compo

C O. 16 0.20

Mn 1. 08 0.45

p 0.010 0.012

S 0.019 0.027

Si 0.25 o. 10

Cu 0.28 ----

V 0.061 ----

TABLE 2 Chemical and Physical Properties

of Girder Plates

20

3 / 16" (A3 6)

0.197

40.4

54.7

28

0.20

0.45

0.006

0.014

----

-- --

----

Stress Level (ksi) Cycles to

Specimen Nos. Min. Max. Initial Crack

31030 H io 30 1,842,000

31530 H (1) 15 30 -------

31530 H (2) 15 30 -------

41030 H 10 30 2,041,000

40530 H (1) 5 30 888,000

40530 H (2) 5 30 862,000

TABLE 3 Summary of Fatigue-Test Results

Remarks

Second crack at 1,867,000 cycles

Run out; Stopped at 2,941,000 cycles

Run out; Stopped at 2,360,000 cycles

Two cracks were found at 862,000 cycles

N .....

22

F r----+ 20-30

F 15-30

Nr 15-30

10-30

F r----+ 10- 30

NF 5-30

NF~ 5- 30

F I FATIGUE CRACK 8

NF I NO CRACK WITHIN 2xl0 CYCLES

FIG. I TESTING SEQUENCE

~""~ ~

~""""""""~,~~"""~. ~ ~

f"""PULSATING JACKS.

FLG. PL. S"xI/2"(ASTM A 47.,

A _ ~~

~ TEST ISECTI0-l l,\B PL. 3/16';1/4"(ASTM A361'N

1~ - - ~~

( ) .,

£1'

I J I I I I

I I I I ! I

" 2'-6" 2'-6"

~.,

-.J \ ,

"--. 3/16"1\. -0

I V -It)

2 (Typ.) ~ , ~ ~ ~

2'-6" 1'-6" 3'-0" 3'-0" 1'-6" 2'-6" 2'-6" 2'-6" 6"

25!.0"

I

,II I I I I I I , I I I

I I ---~---==-::7---------~~~---·-~-----~-~:,=t :~ III ~l ::

-------- LATERAL SUPPORT / .

FIG. 2 SPECIMEN AND TEST SET- UP.

I I

-CD Ii)

-- /I -"

-It)

@)

N

I I I 12 @ 3" = 3611

FIG.3 LOCATIONS OF LATERAL WEB DEFLECTIONS MEASUREMENTS.

/

FIG.4 MOVEABLE HEAD DIAL RIG

24

p (k)

70

60

50

40

30

20

10

25

/

THEORY

o ~~ __ ~ ______ ~ ______ ~ ____ ~ ______ ~ ____ ~ o 100 200 300 400 500 600

CENTERLINE DEFLECTION ( IN. X 101)

FIG. 5 LOAD Vs CENTERLINE DEFLECTION (41030 H)

o 0-1 0-2 I I I I I

SCALE CINCHES)

o ... rmax

Pmin -....--P=O

41030H R ..!!!.£! = 0·864

Per

P=O

31030H P max = '.521 Per

FIG. 6 WEB DEFLECTIONS (CENTERLINE OF TEST PANEL)

26

P (k)

70

60

50

40

30

20

10

o

THEORETICAL WEB STRAIN.

THEORETICAL FLANGE STRAIN

WEB

27

/~!---r---' COMPRESSION FLANGE

~~.,c=.=====-I TENSION FLANGE

rr T STRAIN GAGE LOCATIONS

200 400 600 800 1000 E (}L in.lin. )

FIG. 7 LOAD Vs STRAIN (41030 H)

1"

""":

28

~ ~

..-:~

I (1,868,000) CYCLES

(1,574,000)

I( 1,842,000) CYCLES I

FIG. 8 CRACK LOCATIONS AND CYCLES TO CRACK

INITIATION FOR SPECIMEN 31030 H

! ~

.::

~1.890.000' 2P4I,OOO}

I (2,041,000) CYCLES I (IN WELD) I

FIG. 9 CRACK LOCATIONS AND CYCLES TO CRACK

INITIATION FOR SPECIMEN 41030 H

.,;:::.

.. ~

29

~ ~

-?

807,000) ( 888,000) CYCLES ,

(577,OOOj Ii I

(820,000)

FIG. 10 CRACK LOCATIONS AND CYCLES TO CRACK

INITIATION FOR SPECIMEN 40530 H (I)

~ "

J

c. p

(862,000)

(862,000) CYCLES) I (862,000) CYCLES

FIG. II CRACK LOCATIONS AND CYCLES TO CRACK

INITIATION FOR SPECIMEN 40530 H (2)

-.. lC -I&.. I&..

30

~ 25 :::> o 0:: ...., Z ...., I&.. !:!: 20 I-m

~ ~ Z <t 0:: 15

~ ...., 0::

Ii;

10

"" W FROM FIG. 8·46 OF REF .. 5 ...... PLATE WITH FILLET

~ tWELOEO ATTACHMENT.

~ .........

~ ~ ~ o 3/16" SPECIMENS

A V4" SPECIMENS -........ .... e.... RUNOUT. e-- ')olio-

4X10!5 Ix 106 2xl06

CYCLES TO FIRST CRACK

FIG 12. STRESS RANGE - CYCLES TO FIRST CRACK.

-

-

-

-

...., C) Z <t

25 0::,

m m ...., 0:: I­m

20...., C) Z <t ...J I&..

15

31

FIG. f3 TENSION SPECIMEN WITH FILLET­WELDED ATTACHMENT (5)

5

.: 10 .. LLI i 15 ct

32

x - FAILURE

o - NO FAILURE

31530 H (tl 31530 H (2)

41030 H a: ~ 20~~n~--------------~r--------------------

~ 31030H ~ 25~--------------------------4~~----~~~------­I- 40530 H (J) 40530 H (2) o 30~------------------------------------------

ORDER OF TESTING ~

FIG. 14 5T AIRCASE ILLUSTRATION OF RESULTS

-(f) ~

z -

50

40

LaJ 30 (!) z « 0::

(f) (f)

~ 20 I-C/)

10

o

'" " 0 A 441 FLANGE a A 36 WEB

~ ~ REFERENCE 3 UNDER BENDING

0+- RUN OUT

~'-" '" "'" (J ~

........... i'-.... j'.... r.:\ h

~ 1-

~ .. -----

2x10~ 3x10~

CYCLES

FIG. 15 S- N CURVE FOR PREVIOUS TESTS

-

r--

w w